The world’s leading publication for data science, AI, and ML professionals.

Suggestions For Professional Dashboards

How to build effective data visualizations for Business Executives

Office Hours

Photo by Lukas Blazek on Unsplash
Photo by Lukas Blazek on Unsplash

In the last years, with the wide accessibility of big amounts of data, getting insights from the available information has become an important step in any decision-making process.

Especially when data are Big Data, to better understand what’s going on, you need to organize, summarize, and visualize information.

But even with visualizations, all of your efforts could be vain if you don’t follow some simple guidelines to make your Dashboards effective.

Of course, the purpose and the target of your analytics are the first things to consider when you design a new dashboard.

This is why, in this article, I would like to focus on Professional Dashboards for Business Executives with the purpose to represent in an effective way how the business is going in terms of sales, costs, income, clients, and whatever is important for the first line of your managers.

Know your audience

After defining the purpose and the target, you need to focus not only on the role the managers play in the company but also in their preferences and tastes.

You need to know for example:

  • What is their focus and, above all, what is their focus right now?
  • Are the managers who will receive the dashboards likely to appreciate fancy and modern representations of data (steam graph, sunburst charts, chord diagrams, etc.), or is better to go with a bar chart?
  • Do they love tables?
  • Do they prefer to receive a PDF or a link to a web dashboard?

Build a Prototype

There is nothing worse than working hard and misfiring. Moreover, once you built a dashboard you will feel very protective of it and you will have a hard time accepting any negative comment.

So, before you work weeks on a potential dashboard, taking care of every detail, better to build first a prototype and test it with colleagues.

The prototype should reproduce the main functionality of the dashboard and, in most cases, it may contain sample but realistic data. The people who will look at it should have a general idea of the subject of the report but should not be super-expert.

Listening, evaluating, and applying suggestions at this stage will be easier and faster. In my experience, some of the most appreciated analysis or features could easily come from these comments.

Keep it simple

In my experience, high-level managers don’t have enough time and their attention is usually directed to specific projects or topics.

Therefore, you need to keep in mind that the messages you want to convey should be easily understandable from the dashboard you are building.

Trying to fit a lot of information in one dashboard could result in a mess. So, probably better to get rid of the excess.

Focus on specific subjects could be referenced in the dashboard with links to other dashboards or other pages.

Probably your managers won’t have enough time to understand fancy charts they have not seen before, so probably you should stick with classic charts they already know (bar chart, scatter plot, line chart, etc.). But again, this depends on the managers.

Removing the excess

When you have to remove the excess there are some usual suspects to look for.

  • Axis: are they necessary or the chart have enough information? Usually in a bar chart (with labels on the bars) axis are expendable while in a scatterplot axis are necessary.
  • Values: Can you scale your values to avoid big numbers? The decimal values are necessary? 20,000,000 is far worse than 20bln and adding a decimal point after a number with 4 or 5 significant figures is generally redundant.
  • Additional information. Probably, you’ll need to add some information to your dashboard (how to read a chart, specifics of how a measure is calculated, etc.). My suggestion is to keep in the main dashboard only the more relevant notes and move what you can into footnotes, appendix, or a hideable layer. In any case, do not write too much, this is not a textbook.
  • Double Axis. Often, the double-axis is used to represent two pieces of information in the same chart. In my opinion, one should use the double-axis only to show the presence (or absence) of correlations.

Indicators and KPIs

It could be useful to add the most important indicators and KPIs in the uppermost part of the dashboard with a big font size. That’s the part where they will start reading the report and, for this reason, that’s where they will look for the most important numbers.

Surprise

Add some surprise. That could be some interaction with the chart the managers don’t expect.

If your visualization tools have this kind of feature, you can add an additional analysis or a focus when you click on a specific point of the dashboard. That is could be also a way to remove an analysis from the dashboard and show it on demand.

Keep it clean

There some things that contribute to creating a mess:

  • using many font sizes in the same dashboards. There could be some exceptions but they should be justified (for example big numbers representing KPI at the top of the dashboard);
  • using many font colors in the same dashboards. Again exceptions must be justified;
  • not aligned shapes;
  • using the same type of chart to represent different things.

Maybe you don’t realize it but I assure you that all these things contribute to making readers uncomfortable (and even most of them won’t know where this discomfort comes from).

The font size should be big enough to be easily visible from the devices with which they will access the dashboard. Be maniac in aligning charts and measuring sizes.

If you have brand identity guidelines (where is specified which colors and which font you can use) follow it rigorously.

Moreover, if your managers already receive dashboards from other departments, try to keep the format uniform to avoid confusion. For example, if your managers are used to having filters on the left side is disadvantageous to move them to another side (they will lose time and patience looking for the filters).

Keep it Fast

Ok, now you have a dashboard that has passed a robust "corridor test" but when you open it takes some minutes to load all the charts. Too much for a manager. I would recommend staying below 10 seconds.

There’s a lot of ways to reduce the loading-time, some depend on the visualization tool you are using (Tableau, Power BI, etc.). But most are quite general:

  1. first of all, if your dashboard contains data from many sources that could result in a multiplication of the time to refresh. So, try to understand if all of your sources are necessary (get rid of the excess, again)
  2. Most of your dashboard will consist of aggregating and filtering data. In this case, you can create new tables in your databases that contain only useful data and keep the original tables (with all the data) for Exploratory Data Analysis.

Sometimes, data sources could be blended together and that could save time.

Moreover, if your data are time-related, it could be the case to filter out some old dates. Do you really need daily data for the past 3 years? Or maybe you can settle for daily data for the last 3 months and end-of-month data for the past?

Summing up

A well-designed and tailor-made dashboard could make the difference between reaching your audience and help them to make the right decision or to be ignored throwing away a lot of hard work.

In this article, focused on professional dashboards for executives, I shared some of the things to keep in mind when building a new visualization:

  • one size doesn’t fit all, so better to focus on your main readers;
  • build a prototype and test it with as many people as you can;
  • throw away what you don’t need, make visible what you really need;
  • take care of the details, they can make the difference.

Follow this suggestion and your readers will love your dashboards!


Related Articles